Wegdekmeten.nl is the leading company in the Netherlands in the field of surveying road surface, signs and markings. With special camera cars they make recordings of every meter of road surface, all the white road markings, all traffic signs, every light pole and every centimeter of guardrail alongside the Dutch roads. Commissioned by infrastructure contractors or by local, provincial and national governments, Wegdekmeten.nl records and processes hundreds of thousands of images covering many thousands of kilometers of infrastructure every year. Until recently, the images were analyzed manually by road inspectors, looking for damage, skew or other deviations.
If the execution of a task requires human intelligence, but the handling of the task is inhumanly large, the deployment of artificial intelligence is valuable. It is therefore not surprising that many organizations have tried to use A.I. for detecting asphalt damage. However, until now this has proven to be an almost impossible task due to the astronomical number of variations in damage examples. Where others have attempted (and failed) to train neural networks with deep learning to classify damages, Bolesian has opted for a different approach with our platform Sentinizer™. With our ‘computer vision’ module we remove noise from images and reduce the image information to what is essential. We have interviewed inspectors, made their knowledge explicit and captured their knowledge in special classification models. In this way we have created ‘synthetic brains’ that enable the Sentinizer bots to recognize and classify all deviations in the recorded visual data.
An experienced inspector can inspect and diagnose a maximum of about 20 kilometers of road surface images a day. Bolesian’s Sentinizer™ processes a multiple and does so 24/7, taking this processing to a whole new and unprecedented level. A truly remarkable performance that accelerates the business of Wegdekmeten.nl and demonstrates the power of artificial intelligence when applied by Bolesian’s specialists.